Fertility and Birth Outcomes in Women With Epilepsy Seeking Pregnancy

Key Points

Question  Do women with epilepsy without a prior diagnosis of infertility or related disorders have the same likelihood of achieving pregnancy as their peers without epilepsy?

Findings  This multicenter cohort study of 89 women with epilepsy and 108 control women found no difference in the proportion of women who achieved pregnancy less than 12 months after enrollment in the study. Among women with epilepsy, 54 (60.7%) achieved pregnancy vs 65 (60.2%) among control women.

Meaning  Women with epilepsy without a history of infertility or related disorders seeking pregnancy had similar likelihood of achieving pregnancy compared with their peers without epilepsy.

Importance  Prior studies report lower birth rates for women with epilepsy (WWE) but have been unable to differentiate between biological and social contributions. To our knowledge, we do not have data to inform WWE seeking pregnancy if their likelihood of achieving pregnancy is biologically reduced compared with their peers.

Objective  To determine if WWE without a prior diagnosis of infertility or related disorders are as likely to achieve pregnancy within 12 months as their peers without epilepsy.

Design, Setting, and Participants  The Women With Epilepsy: Pregnancy Outcomes and Deliveries study is an observational cohort study comparing fertility in WWE with fertility in control women (CW) without epilepsy. Participants were enrolled at 4 academic medical centers and observed up to 21 months from November 2010 to May 2015. Women seeking pregnancy aged 18 to 40 years were enrolled within 6 months of discontinuing contraception. Exclusion criteria included tobacco use and a prior diagnosis of infertility or disorders that lower fertility. Eighteen WWE and 47 CW declined the study, and 40 WWE and 170 CW did not meet study criteria. The Women With Epilepsy: Pregnancy Outcomes and Deliveries electronic diary app was used to capture data on medications, seizures, sexual activity, and menses. Data were analyzed from November 2015 to June 2017.

Main Outcomes and Measures  The primary outcome was proportion of women who achieved pregnancy within 12 months after enrollment. Secondary outcomes were time to pregnancy using a proportional hazard model, pregnancy outcomes, sexual activity, ovulatory rates, and analysis of epilepsy factors in WWE. All outcomes were planned prior to data collection except for time to pregnancy.

Results  Of the 197 women included in the study, 142 (72.1%) were white, and the mean (SD) age was 31.9 (3.5) years among the 89 WWE and 31.1 (4.2) among the 108 CW. Among 89 WWE, 54 (60.7%) achieved pregnancy vs 65 (60.2%) among 108 CW. Median time to pregnancy was no different between the groups after controlling for key covariates (WWE: median, 6.0 months; 95% CI, 3.8-10.1; CW: median, 9.0 months; 95% CI, 6.5-11.2; P = .30). Sexual activity and ovulatory rates were similar in WWE and CW. Forty-four of 54 pregnancies (81.5%) in WWE and 53 of 65 pregnancies (81.5%) in CW resulted in live births. No epilepsy factors were significant.

Conclusions and Relevance  Women with epilepsy seeking pregnancy without prior known infertility or related disorders have similar likelihood of achieving pregnancy, time to pregnancy, and live birth rates compared with their peers without epilepsy.

Approximately 12.5 million women of childbearing age worldwide have epilepsy.¹ Most studies suggest birth rates in women with epilepsy (WWE) to be 37% to 88% of other groups.^2,3 Conversely, the Northern Finland Birth Cohort⁴ reported the number of children born to WWE did not differ from the reference group overall, although epilepsy not in remission was associated with fewer children. None of these studies collected information about the desire or attempts to achieve pregnancy. Birth rates could be lower in WWE because of social factors (eg, lower marriage rates or lower rates of seeking pregnancy) and/or biological factors (eg, decreased ovulatory rates). A UK survey⁵ reported 33% of WWE respondents were not considering having children because of their epilepsy. Neurologists do not have the information needed to counsel their female patients who desire pregnancy whether they are as likely to achieve pregnancy and to have a live birth as their peers.

We designed this study to determine whether WWE without a preexisting diagnosis of infertility or associated disorders who are following their clinically determined treatment regimen differ from women without epilepsy when attempting to achieve pregnancy. The primary aim was the proportion of women achieving pregnancy within 1 year. Secondary aims were comparisons between WWE and women without epilepsy for (1) live birth rates, (2) time to achieve pregnancy, (3) sexual activity rates, and (4) ovulatory rates. An additional secondary aim was to explore seizure and medication factors within WWE that could affect fertility.

To our knowledge, no prior studies prospectively compared WWE with women without epilepsy attempting to conceive. Our study approach excluded women with known infertility diagnosis or diagnoses associated with reproductive disorders (eg, polycystic ovary syndrome [PCOS]). This study specifically addressed the woman who arrives in clinic with no reason to suspect infertility other than an epilepsy diagnosis.

We enrolled WWE and control women (CW) without epilepsy seeking pregnancy at 4 academic medical centers in the United States: Brigham and Women’s Hospital, Boston, Massachusetts, New York University Medical Center, New York, Northwell Health (formerly North Shore–Long Island Jewish Health System), Great Neck, New York, and the University of Pittsburgh, Pittsburgh, Pennsylvania. We recruited WWE and CW from flyers and brochures placed in neurology and gynecology clinics at the clinical sites and in local restaurants, churches, childcare centers, hospital research bulletins, email listserves, and online postings. Institutional review board approvals were obtained from Partners HealthCare System for Brigham and Women’s Hospital, New York University, Northwell Health Institutional Review Board (formerly North-Shore Long Island Jewish Hospital Institutional Review Board), and University of Pittsburgh Medical Center. All participants provided written informed consent. The study was registered on clinicaltrials.gov (NCT01259310).

The rate of pregnancy in WWE was hypothesized to be two-thirds the rate for CW. Pregnancy rate by 12 months after enrollment was chosen as the primary end point given its clinical relevance and consistency with the World Health Organization criteria for a diagnosis of infertility.⁶ We used published 3-month pregnancy rates to conservatively calculate sample size for pregnancy rates by 12 months after enrollment. The population-based estimate of pregnancy rate over 3 menstrual cycles is approximately 55%⁷; thus, we estimated that the rate among WWE was expected to be approximately 36%. Based on a 2-tailed comparison of rates, we estimated that a sample size of 100 women in each group would be needed to provide 79% power to detect this difference at a significance level of .05. To account for dropout, we planned to recruit a total of 110 women in each group.

Women with epilepsy and CW were included if they were aged 18 to 40 years, expressed a desire to conceive, had a steady male partner, and had discontinued any method of contraception within the previous 6 months or planned to do so within 6 months of providing informed consent. The study protocol commenced after contraception was discontinued in the latter case. For intramuscular medroxyprogesterone acetate, a woman could not be enrolled until 6 months after last injection and menses were regular. Exclusion criteria were a current attempt to achieve pregnancy for 6 or more months; demonstrated infertility with the same partner anytime in the past⁶; breastfeeding; diagnoses of infertility, PCOS, severe endometriosis, menopause, untreated thyroid disease, nipple discharge, hyperprolactinemia, or other pituitary disease; smoking more than 10 cigarettes per day; or planning to use medication to conceive or a sperm donor. Women were also excluded if their male partner had a history of abnormal semen analysis results, vasectomy reversal, or male factor infertility.

At the enrollment visit, demographic information and medical history were obtained, including results of ancillary studies, seizure types, and epilepsy treatments; a fertility questionnaire was administered; physical and neurological examinations were performed; and seizures, etiology, and epilepsy syndromes were classified using the National Institute of Neurological Disorders and Stroke Epilepsy Common Data Elements.⁸ Irody Inc designed an electronic daily diary app and the cloud data system (the Women With Epilepsy: Pregnancy Outcomes and Deliveries [WEPOD] study diary) for participants (Figure 1). We installed the app on a provided iPod Touch (Apple) or personal smartphone and instructed all participants on how to track menses and sexual intercourse as well as medication use and seizures for the epilepsy group. Participants could also enter data via a web-based interface or choose to use a paper diary.

Participants had monthly visits in person or via telephone for review of their diary. Participants were instructed to conduct a home pregnancy test if menses did not begin by day 35 of their cycle. If positive, they had an initial pregnancy visit and a 1-month postpartum visit. Pregnancy losses (ie, ectopic pregnancy, miscarriage, fetal death, and abortions) were included in the positive pregnancy group. Participants were asked to sign a release to obtain medical records from their health care team and the delivery hospital to verify pregnancy outcomes. Each participant contributed only 1 pregnancy to these analyses.

For each participant, time to pregnancy was calculated as the interval between the end contraception date and 14 days after the first day of the last menstrual period. Women who did not achieve pregnancy within 12 months of beginning the protocol or who dropped out without becoming pregnant were considered censored in the analysis.

For each participant, months of daily electronic diary data with more than 80% of days with tracked sexual activity were analyzed. Sexual activity rate was calculated as the number of days with at least 1 episode of sexual intercourse divided by the number of days in eligible months prior to pregnancy diagnosis or study end.

We obtained venous blood for serum samples within 1 day of menstrual day 21 for up to 2 cycles. Samples collected within 14 days of the next menstrual period were considered in the luteal phase and included in this analysis. Progesterone level was measured in the Brigham Research Assay Core in Boston, Massachusetts, via access chemiluminescent immunoassay in batched samples. Intra-assay variation was reported as 6.11% to 11.19% and interassay variation as 6.59% to 9.57%. The menstrual cycle was determined as ovulatory if the progesterone level was 3.0 ng/mL or greater.

We summarized all categorical variables by proportions and continuous variables by means, medians, standard deviations, and interquartile ranges. The primary analysis used a χ² test to compare the proportion of women achieving pregnancy between groups. In secondary analysis, we used Kaplan-Meier curves, the log-rank test, and Cox proportional hazards models to assess the association between time to pregnancy and epilepsy status while controlling for baseline characteristics. All factors significant at the .20 level in univariate analyses were considered in multivariable models, as were covariates that are consistently shown to affect fertility (ie, age and body mass index). The importance of each variable was assessed through likelihood ratio tests at the .10 significance level. The process of removing variables, refitting, and retesting was continued until all variables in the model had a P value of .05 or less.

Pregnancy outcomes were analyzed as proportions of miscarriages to live births across groups using a χ² test. Rates of sexual activity were estimated and compared between the WWE and CW using Poisson regression. Ovulatory status was analyzed using logistic regression and Generalized Estimating Equations to account for the intraparticipant correlation.

Separate analyses assessed the additional contribution of epilepsy factors in the epilepsy cohort only, including seizures in the 9 months prior to enrollment (yes/no) and enzyme-inducing antiepileptic drug (EIAED) use. Carbamazepine, phenytoin, phenobarbital, primidone, and oxcarbazepine were categorized as EIAEDs. Statistical significance was set at P < .05, and all P values were 2-tailed. All analyses used SAS statistical software version 9.2 (SAS Institute).

Because WWE and CW were recruited from different sources out of necessity, a higher percentage of WWE signed informed consent prior to stopping contraception. To rule out any obscuration of a difference between the groups in time to pregnancy, a sensitivity analysis was performed by removing the participants enrolled prior to stopping contraception.

Race and ethnicity were self-reported. National survey data of infertility in the United States examined racial and ethnic differences. The only category that differed was non-Hispanic black women, who were more likely to be infertile than non-Hispanic white women.⁹ As a sensitivity analysis, we repeated the analysis after removing non-Hispanic black women.

Prior studies demonstrate that the probability of conception is highest in the first cycle without contraception and drops off with each subsequent cycle.^7,10,11 Because participants could enroll until 6 months after contraception discontinuation, we compared the proportion of WWE and CW enrolling less than 3 months and 3 to 6 months after discontinuing contraception.

We enrolled 89 WWE and 108 CW (Table 1) from November 2010 through February 2014. Primary reasons for not enrolling in the WEPOD study after initial inquiry are listed in Table 2. Recruitment was stopped below the planned 100 WWE because of slow enrollment but was partially compensated by enrollment of additional CW. There were 18 early discontinuations from the epilepsy group and 15 from the control group, who were censored at the time of dropout.

For the primary aim, we found no differences between the epilepsy and control groups in the proportion who achieved pregnancy (WWE, 60.7% [54 of 89]; CW, 60.2% [65 of 108]; χ² = 0.28; P = .60). Median time to pregnancy was 6.0 months (95% CI, 3.8-10.1) in WWE and 9.0 months (95% CI, 6.5-11.2) in CW (P = .30) (Figure 2). Women who had features of higher parity, were white, and were married were more likely to achieve pregnancy, and women with a higher age at enrollment were less likely to achieve pregnancy (eTable in the Supplement). Time to pregnancy was no different after controlling for parity, race/ethnicity, marital status, age, and body mass index.

Pregnancy outcomes were similar, with 81.5% of pregnancies producing live births for each group (44 of 54 pregnancies in WWE and 53 of 65 pregnancies in CW). Miscarriage rates were similar, occurring in 8 pregnancies (14.8%) in WWE and in 12 pregnancies (18.5%) in CW. Additionally, of the WWE pregnancies, 1 ectopic pregnancy (1.9%) and 1 therapeutic abortion caused by genetic anomalies (1.9%) occurred.

A total of 145 participants had months with more than 80% daily tracking of sexual activity and were included in the analysis, including 57 of 89 WWE (64.0%) and 88 of 108 CW (81.5%). We found no difference in sexual activity rates between WWE and CW (Figure 3).

A total of 243 cycles were included in the analysis. Six cycles in WWE and 17 cycles in CW were excluded because of the progesterone sample being drawn more than 14 days prior to the next menstrual period. In WWE, 92 of 104 cycles (88.5%) were ovulatory. In CW, 113 of 139 cycles (81.3%) were ovulatory. The relation between ovulation and study group was not significant.

In the analysis of the epilepsy cohort, 2 participants were removed because they were not taking antiepileptic drugs (AEDs) (Table 1). All other WWE continued to use their prescribed AEDs. Presence of active seizure in the prior 9 months did not influence the likelihood of pregnancy. Women receiving an EIAED were less likely to achieve pregnancy compared with women taking other AEDs (hazard ratio, 0.456; 95% CI, 0.193-1.081).

Results of sensitivity analyses were similar to initial analyses. After removing women who stopped birth control after enrollment (14 WWE and 2 CW), adjusted times to pregnancy were a median of 7.3 months (95% CI, 3.8-12.2) in WWE and 9.0 months (95% CI, 6.5-11.2) in CW (P = .75). One non-Hispanic black WWE and 16 non-Hispanic black CW were removed from the time to pregnancy analysis; findings did not change. There was no difference between groups for duration not taking birth control prior to enrollment, with 74 of 89 WWE (83.1%) and 87 of 108 CW (80.6%) enrolled within 3 months of stopping birth control and the remainder enrolled 3 to 6 months after stopping birth control.

In this prospective study of WWE seeking pregnancy, the first such study, to our knowledge, that included controls, we provide substantial data to support that WWE have similar pregnancy rates and time to achieve pregnancy as women without epilepsy of similar age. These findings are specific to women who do not have established diagnoses of infertility or associated disorders. Prior research has suggested WWE are at higher risk of reproductive endocrine dysfunction,¹² and PCOS occurs more frequently in WWE, especially with valproate use in idiopathic generalized epilepsy syndromes.^13-16 Therefore, we excluded women with PCOS to be able to determine if women without known PCOS have impaired fertility. Our findings are directly applicable to the more common clinical scenario of WWE without a preexisting diagnosis of infertility or associated disorders seeking information from their physicians when they decide they want to become pregnant.

The US Centers for Disease Control and Prevention report approximately 6% of married women in the United States are infertile,¹⁷ although other estimates are higher.¹⁸ The lower 12-month pregnancy rates in these groups of WWE and CW can be attributed to the fact that we enrolled women up to 6 months after birth control cessation. The probability of conception is highest in the first cycle without contraception and drops with each subsequent cycle.^7,10,11,19 The proportion of WWE and CW enrolling less than 3 months and 3 to 6 months after discontinuing contraception did not differ, and thus, this was not likely to bias our data. Additionally, the ages of participants were higher in this study than other studies, which is expected to lower pregnancy rates.^10,19 Furthermore, our reported pregnancy rates do not adjust for the 33 early discontinuations.

A prospective cohort study in India²⁰ reported 3 or more AEDs and use of phenobarbital were associated with higher risk of infertility. Other studies reported that EIAEDs contribute to higher rates of menstrual disorders.^21,22 In the WEPOD study, only 9 WWE (10.1%) were taking AED polytherapy and none were taking phenobarbital, which may explain why the WEPOD study did not similarly show diminished fertility. The WEPOD study only enrolled 16 WWE (18.0%) taking an EIAED, reflective of the practices at the study sites. The hazard ratio for women taking EIAEDs to achieve pregnancy was 0.456 (95% CI, 0.193-1.081) but did not reach statistical significance. Our subgroup analyses in the epilepsy group were secondary and may not have shown differences because of low power. Therefore, the potential influence of EIAEDs on fertility is worth studying in a larger group of women.

Other studies suggest differences in fertility between active vs controlled epilepsy.^4,23 In this cohort, 39 of 87 WWE (44.8%) had seizures within the prior 9 months; we did not find a difference in pregnancy outcomes between these women and seizure-free women.

Because studies reported higher rates of anovulatory cycles and sexual dysfunction in WWE,^24-26 we explored these 2 important factors for pregnancy. We did not find differences in ovulation or sexual activity rates between the 2 groups. However, when the goal is to conceive, libido plays a less important role in rates of sexual activity, so our data cannot provide direct inferences regarding libido in WWE.

The number of live births and miscarriages did not differ between the 2 groups in the WEPOD study, and rates were within expectations for women who track prospectively prior to conception.^27,28 Although some studies suggest WWE are at higher risk for miscarriage,²⁹ to our knowledge, no prior studies exclusively enrolled women prior to pregnancy and likely missed capture of early miscarriages. The International Registry of Anti-Epileptic Drugs and Pregnancy study³⁰ reported increased miscarriage risk with antiepileptic polytherapy, and it is possible that our study had too few women taking polytherapy to detect this difference. However, all miscarriages in this study occurred in women on AED monotherapy. Our findings are similar to another study³¹ that found no difference in miscarriage rates in WWE (18%) compared with wives of men with epilepsy.

A unique characteristic of the WEPOD cohort is that all women planned their pregnancy in advance, differing from pregnancy registry or population studies. A strength of our study was use of a digital diary to capture medication use, seizures, sexual activity, and menstrual bleeding. This, coupled with progesterone level measurements to confirm ovulation, allowed us to address potential fertility factors that could not be studied otherwise. Women were very compliant with the diary, as has been published elsewhere.³²

There were limitations in our study. This study does not answer if there are differences in rates of infertility or PCOS, as women with these diagnoses were excluded from both groups. We were not able to prospectively track data from the date of stopping contraception in most participants. Some women may have achieved pregnancy before they could be recruited, making pregnancy rates lower than expected. We accounted for all anticipated confounders associated with fertility. Our analysis showed expected associations (ie, age and gravidity) and some unanticipated associations (ie, race and marital status). Although inclusion criteria required a stable male partner, it is possible that being married was associated with shorter time to pregnancy because of greater availability for intercourse and/or stronger desire to achieve pregnancy expeditiously. In the enrolled women, higher rates of marriage were present in white participants (90.1% [128 of 142]) vs other races (58.2% [32 of 55]) and could have contributed to the unexpected finding of an association of white race with shorter time to pregnancy. The epilepsy group had greater proportions of white participants and married participants than the control group. However, because these covariates were incorporated into the model, it is unlikely any of these differences influenced the primary findings of no differences in pregnancy rates or time to pregnancy. The possibility remains that additional unmeasured differences between WWE and CW were not accounted for in the model and could have influenced the results, especially because different sources of recruitment were necessary.

In this prospective cohort study of women without an a priori diagnosis of infertility or related disorder, WWE seeking pregnancy experienced similar pregnancy rates compared with their peers without epilepsy. Secondary analyses also showed similar times to achieve pregnancy and pregnancy outcomes. Although this study design cannot exclude an increased risk for impaired fertility across all WWE, our results should reassure WWE without a prior diagnosis of infertility or related disorders and the clinicians who care for them when planning pregnancy.

Accepted for Publication: January 25, 2018.

Corresponding Author: Page B. Pennell, MD, Brigham and Women’s Hospital, 75 Francis St, 60FE-4052, Boston, MA 02115 (ppennell@bwh.harvard.edu).

Published Online: April 30, 2018. doi:10.1001/jamaneurol.2018.0646

Author Contributions: Dr Pennell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Pennell, French, and Harden all contributed equally to the research study and manuscript.

Study concept and design: Pennell, French, Harden, Davis.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Pennell, French, Harden.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Harden, Bagiella, Andreopoulos, Lau, Barnard.

Obtained funding: Pennell, French, Harden.

Administrative, technical, or material support: Pennell, Harden, Lau, Llewellyn, Allien.

Study supervision: Pennell, French, Harden.

Conflict of Interest Disclosures: Drs Pennell, French, Harden, and Davis and Mss Andreopoulos, Lau, Llewellyn, Barnard, and Allien have received grants from Milken Family Foundation, the Epilepsy Therapy Project, and the Epilepsy Foundation during the conduct of the study. Dr Pennell has received grants from the National Institutes of Health during the conduct of the study as well as honoraria and travel reimbursements from the American Academy of Neurology, American Epilepsy Society, the National Institutes of Health, academic universities, the Peru Ministry of Health, and the Indian Academy of Neurology. Dr French has received salary support to New York University from the Epilepsy Foundation and for consulting work on behalf of the Epilepsy Study Consortium for Acorda, Adamas, Alexza, Anavex, Axcella Health, Biogen, BioPharm Solutions, Cavion, Cerecor, Concert Pharmaceuticals, Engage, Eisai, GlaxoSmithKline, GW Pharma, Marinus, Neurelis, Novartis, Pfizer, Otsuka, Ovid, Sage Therapeutics, SK Life Sciences, Sunovion, Takeda, UCB, Upsher Smith, Xenon Pharmaceuticals, Zogenix, and Zynerba. Dr French has also received research grants from Acorda, Alexza, Eisai Medical Research, Lewis County General Hospital, Lundbeck, Pfizer, SK Life Sciences, Sunovion, Takeda, and UCB as well as grants from the Epilepsy Research Foundation, Epilepsy Study Consortium, Epilepsy Therapy Project, and National Institute of Neurological Disorders and Stroke. Dr French serves on the scientific advisory boards of Ovid, Sage Therapeutics, and Blackfynn, and she serves on the editorial board of Lancet Neurology, Neurology Today, and Epileptic Disorders. She has received travel reimbursement related to research, advisory meetings, or presentation of results at scientific meetings from the Epilepsy Study Consortium, the Epilepsy Foundation, Eisai, GW Pharma, Marinus, Novartis, Pfizer, Sage, SK Life Sciences, Takeda, UCB, Upsher-Smith, Zogenix, and Zynerba. No other disclosures were reported.

Funding/Support: The Milken Family Foundation, Epilepsy Therapy Project, and Epilepsy Foundation provided financial support for the Women With Epilepsy: Pregnancy Outcomes and Deliveries study. Study data were collected and managed using REDCap (Research Electronic Data Capture) tools hosted at Brigham and Women’s Hospital.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank Anto Bagic, MD, PhD (University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania), and Sean Hwang, MD (Northwell Health, Great Neck, New York), for supervisory support of research staff and data collection. They did not receive compensation for their work, and they gave permission to include their names in the acknowledgment section.

Additional Information: REDCap is a secure, web-based application designed to support data capture for research studies, providing (1) an intuitive interface for validated data entry, (2) audit trails for tracking data manipulation and export procedures, (3) automated export procedures for seamless data downloads to common statistical packages and (4) procedures for importing data from external sources.